Audio SSML tag does not work using progressive response

Question

question

NewUser-eda1856a-03d7-409a-9e93-85078d2f0bca asked · Dec 09, 2018 at 08:27 AM

Audio SSML tag does not work using progressive response

Using VoicePlayer.Speak directive for a progressive response on python 2.75, ask_sdk_core 1.5.0 and ask_sdk_model 1.4.0.

We send the request as usual to api_endpoint+'/v1/directives', using Beared Authorization + Api Access token, etc. and It works with a simple text message (as long as we don't enclosed it in <speak> </speak> tags). It plays the message in a lower volume level, but it works.

Using <speak> tags (with text or audio) the request always returns a 400 error.

If we don't use the speak tags, it returns a valid 204 status code, playing the text and ignoring the <audio src> tag.

We tried using audio files of less than 30 sec. stored in S3 and also some sound bank examples with the same frustrating result, it ignore the audio tags or return an error using the <speak> tags.

<code><audio src='soundbank://soundlibrary/animals/amzn_sfx_bear_groan_roar_01'/>

Other SSML tags like <say-as interpret-as='interjection'>fu</say-as>" works perfectly.

It seams the problem is focused only on the audio tag. Any workaround?

Thanks.

audio python text-to-speech progressiveresponse

0

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

9 Answers

Answer 1 · 2020-04-07T04:39:32Z

Bogdan Kulbida answered · Apr 07 at 04:39 AM

Any updates on this issue?

0 · Show 1 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

barry@amazon ♦♦ · Apr 07 at 09:46 AM

Hi @Bogdan Kulbida and thanks for the message.

At the time of this reply, this is still not supported.

Regards,
Barry

0 · Share ·

Answer 2 · 2019-11-27T15:47:04Z

MakeBots answered · Nov 27, 2019 at 03:47 PM

Its now November 2019 and I still have this issue that audio wrapped in <speak> does not work but plain text does work in a progressive response. The answer above were months ago. Is this still being worked on?

0 · Show 2 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

MakeBots · Nov 27, 2019 at 07:35 PM

OK, I've got to the bottom of the problem. Soundbank files do not work at all, but if you snapshort the sound convert to MP3 and convert using ffmpeg as defined in the docs to 48kps and rate 16000, mpeg2 audio and use the <speak> SSML wrapper and host that file yourself it does work. Strange you've created a useful sound library with soundbank that does not work for the most useful use case for sounds. Get your act together guys - this problem has been around for 12 months!!!

1 · Share ·

barry@amazon ♦♦ MakeBots · Dec 02, 2019 at 05:14 PM

Thank you for your deep dive on this @MakeBots.

I understand your frustration but I assure you that this is being tracked internally. Keep an eye open for updates.

Regards,
Barry

0 · Share ·

Answer 3 · 2019-05-21T21:39:22Z

viscusog answered · May 21, 2019 at 09:39 PM

What's the status?

0 · Hide 1 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

barry@amazon ♦♦ · May 23, 2019 at 02:52 PM

Hi @viscusog,

I has not been implemented yet. If you could share your business/CX case, I can try and push it internally.

Regards,
Barry

0 · Share ·

Answer 4 · 2019-04-30T04:06:47Z

D. Young answered · Apr 30, 2019 at 04:06 AM

@barry@amazon - has the bug been fixed?

"A bug was identified that is preventing the Progressive Response to play audio samples from soundbank. A fix is being developed and should be deployed soon. "

0 · Show 1 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

barry@amazon ♦♦ · May 20, 2019 at 04:09 PM

Hello @D. Young and sorry for my late reply.

This is currently in the team backlog, but if you could share your business/CX case, it would help me to push it internally.

Regards,
Barry

0 · Share ·

Answer 5 · 2019-03-06T19:18:46Z

BR-Ryan answered · Mar 06, 2019 at 07:18 PM

Hi Barry!

Have there been any updates on this issue? I am able to play soundbank audio in a response, but not in a VoicePlayer.Speak directive.

I'm trying to use

<audio src='soundbank://soundlibrary/musical/amzn_sfx_electronic_major_chord_01' />

to play in between lists of items.

Thanks!

Ryan

0 · Show 3 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

barry@amazon ♦♦ · Mar 08, 2019 at 12:46 PM

Hello @BR-Ryan and thanks for posting.

This could be happening depending on how you are implementing the Progressive Response.

Regards,
Barry

0 · Share ·

BR-Ryan barry@amazon ♦♦ · Mar 08, 2019 at 03:59 PM

Hi Barry,

Thanks for the response!

I'm implementing like the examples from the Alexa cookbook project. (https://github.com/alexa/alexa-cookbook/tree/master/feature-demos/skill-demo-progressive-response)

If I pass a plain string, she speaks it without any issue. If I pass that same string and append "<audio src='soundbank://soundlibrary/musical/amzn_sfx_electronic_major_chord_01' />", she speaks the text and skips the sound.

I have also tried just passing the sound as the string and she totally skips it.

By comparison, if I pass the same string as a returned response at the end of my intent handler (instead of a progressive response) she speaks the text and plays the sound as expected.

Is there another way of implementing the Progressive Response?

Thanks for your help!

Ryan

0 · Share ·

barry@amazon ♦♦ BR-Ryan · Mar 12, 2019 at 01:59 PM

Hi Ryan,

As mentioned in one of my replies, there is a bug that is preventing Alexa to access sound clips from soundbank. The fix is being worked on and should be deployed soon.

Regards,
Barry

0 · Share ·

Answer 6 · 2019-02-12T03:04:32Z

darianbjohnson answered · Feb 12, 2019 at 03:04 AM

I am not able to get any audio working with progressive response (regardless of it being hosted on S3 or using the audio bank). Are you sure that hosted mp3 files will work with progressive response (assuming that the mp3 meets the audio standards)?

0 · Show 3 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

barry@amazon ♦♦ · Feb 12, 2019 at 04:17 PM

Hello and thank you for your message @darianbjohnson.

The hosted file must be publicly available. Can you please share the link of the file you are trying to use as a source?

Regards,
Barry

0 · Share ·

darianbjohnson barry@amazon ♦♦ · Feb 17, 2019 at 05:21 PM

This is the file I am attempting to use:

https://s3.amazonaws.com/darianchess/static/five_sec_loop.mp3

I've wrapped it in a speak tag as follows:

<speak><audio src="https://s3.amazonaws.com/darianchess/static/five_sec_loop.mp3" /></speak>

0 · Share ·

barry@amazon ♦♦ darianbjohnson · Feb 19, 2019 at 07:52 PM

Hi @darianbjohnson and thanks for the additional information.

I wasn't able to reproduce the issue. Using the URL you provided, I was able to hear the audio. This means that there might be an issue with the way you are building your response. This page has some samples on how to implement the Progressive Response API.

Regards,
Barry

0 · Share ·

Answer 7 · 2018-12-21T10:09:02Z

barry@amazon answered · Dec 21, 2018 at 10:09 AM

Hello and thank you for your patience while we analysed this issue.

A bug was identified that is preventing the Progressive Response to play audio samples from soundbank. A fix is being developed and should be deployed soon.

For audio files hosted in S3, the <audio> tag must be wrapped by the <speak> tag. The following code works:

<speak><audio src="https://s3.amazonaws.com/yourBucket/yourAudioSample.mp3" /></speak>

For <say-as> and <amazon:effect> tags, there is no need to use the <speak> tag.

I hope this helps you.

Regards,
Barry

0 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

Answer 8 · 2018-12-12T18:49:27Z

Anan answered · Dec 12, 2018 at 06:49 PM

I have same problem.

I wasn't sure if it because of Beta hosted skill or the ssml itself.

0 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

Answer 9 · 2018-12-12T16:39:51Z

barry@amazon answered · Dec 12, 2018 at 04:39 PM

Hello and thank you for your message.

I was able to reproduce your issue and have forwarded it to our internal teams for them to check. I will keep you updated of any development.

Regards,
Barry

0 · Show 1 · Share

10 |2000 characters needed characters left characters exceeded

Attachments: Up to 2 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.

NewUser-e3fc3f87-b82b-4202-b1b5-511d20982f57 · Aug 08, 2019 at 03:16 PM

Hey Barry, can you reach out to me on Chime regarding this? bladelee@

0 · Share ·

Audio SSML tag does not work using progressive response

question details

Related Questions